Computational Reverse Engineering Analysis of Scattering Experiments Method for Interpretation of 2D Small-Angle Scattering Profiles (CREASE-2D)

Characterization of structural diversity within soft materials is key for engineering new materials for various applications. Small-angle scattering (SAS) is a widely used characterization technique that provides structural information in soft materials at varying length scales and typically outputs scattered intensity I(q) as a function of the scattered wavevector represented by its magnitude q and azimuthal angle {\theta}. While isotropic structures can be interpreted from azimuthally averaged 1D SAS profile, to understand anisotropic spatial arrangements, one has to interpret the 2D SAS profile, I(q,{\theta}). In this paper, we present a new method called CREASE-2D that interprets I(q,{\theta}) as is and outputs the relevant structural features. CREASE-2D is an extension of the 'computational reverse engineering analysis for scatting experiments' (CREASE) method that has been used successfully to analyze 1D SAS profiles for a variety of soft materials. CREASE uses a genetic algorithm for optimization and a surrogate machine learning (ML) model for fast calculation of 1D 'computed' scattering profiles that are then compared to the experimental 1D scattering profiles during optimization. In CREASE-2D, which goes beyond CREASE in interpretting 2D scattering profiles, we use XGBoost as the surrogate ML model to relate structural features to the I(q,{\theta}) profile. The CREASE-2D workflow identifies the structural features whose computed I(q,{\theta}) profiles match the input experimental I(q,{\theta}). We test the performance of CREASE-2D by using as input a variety of in silico 2D SAS profiles with known structural features and demonstrate that CREASE-2D converges towards their correct structural features. We expect this method will be valuable for materials' researchers who need direct interpretation of 2D scattering profiles to explore structural anisotropy.